Global semi news — Korea, China, Taiwan, the US, and Japan. Government policy, export controls, capex moves, supply-chain shifts, and macro events. AI-classified and tagged with affected tickers. All headlines link back to the originating publisher.
Original: 華碩旗下台智雲打造超級電腦獲選全球前500 囊括台灣公開算力近8成
ASUS (2357) and subsidiary Taiwan Web Service co-built Taiwan supercomputers now exceeding 165 PF, capturing ~78% of Taiwan's TOP500 public compute as the island crossed the 213 PF threshold with 11 systems on the list. NCHC's NANO4 ranks #33 globally at 81.6 PF; TWS is also engaged in Ubilink and Foxconn's (2317) AI supercomputing center, giving ASUS clearer visibility into sovereign-AI cloud compute demand.
Why it matters: Industry/market-data story showing ASUS's growing share of Taiwan sovereign-AI compute and Foxconn's AI datacenter build-out, but no specific contract value, order or earnings catalyst is disclosed.
Original: 三星推进Exynos 2700芯片研发,计划将于2026年末量产 - cnenergynews.cn
Chinese media reports Samsung is advancing its next-gen Exynos 2700 application processor with mass production planned for end-2026, likely on Samsung Foundry's 2nm SF2 node. The framing is neutral-to-skeptical, watching whether Samsung LSI can finally execute after Exynos 2500 delays and reclaim share from Qualcomm in the Galaxy S27 — a key validation point for Samsung's struggling foundry business that also affects TSMC's flagship-SoC dominance.
Why it matters: Exynos 2700 on Samsung 2nm is a key test for Samsung Foundry's recovery and directly affects Qualcomm/TSMC's grip on Galaxy flagship SoC sockets, though no concrete yield or customer data is disclosed.
Open source articleOriginal: 中国芯片出口额暴增110%,是否意味西方“封锁失败”? - BBC
BBC report highlights a 110% surge in China's chip exports, framed by Chinese media as evidence that US-led export controls are backfiring and accelerating domestic substitution. The narrative reinforces bullish sentiment for SMIC/Huawei ecosystem while implying share loss risk for Western and Asian incumbents (Nvidia, TSMC, Samsung, SK Hynix) in the China market.
Why it matters: Macro-level trade data with a clear CN domestic-substitution framing that pressures tracked KR/TW/US incumbents in China, but no single-company catalyst that warrants 'high'.
Original: 〈台股開盤〉記憶體反彈助漲逾700點 但受制5日線反壓
TAIEX opened up over 740 points to 46,785 on Wednesday after Micron's strong earnings and its call for memory shortages through 2027 sent the sector surging, with Nanya Tech (2408) and Winbond (2344) up over 5%. TSMC (2330) gained over 1% to NT$2,420 holding the monthly line, while MediaTek (2454) and Hon Hai (2317) rose nearly 2%, and passive component names including Yageo (2327) also rallied alongside financials.
Why it matters: Broad market open recap tied to Micron's earnings-driven memory rally and a 2027 shortage call — sector-level supply/demand signal rather than a single stock-moving catalyst.
Original: 超過 250 核、時脈破 5GHz!高通首款伺服器 CPU「Dragonfly C1000」正式亮相
Qualcomm announced its first data center server CPU 'Dragonfly C1000' at Investor Day, featuring 250+ Oryon cores at 5GHz+, PCIe Gen7, 2TB/s bandwidth, and 2x power efficiency vs current server CPUs, targeting 2028 mass production. Meta has signed a multi-generation partnership, marking Qualcomm's entry into the $200B server CPU market currently contested by AMD (256-core Venice), Intel (Diamond/Coral Rapids), Nvidia (Vera), and Arm. The chiplet design and HBC memory expansion option signal demand for advanced packaging and HBM-class memory in the 2028 server cycle.
Why it matters: Roadmap/product announcement with 2028 production timeline — no near-term revenue impact, but signals server CPU competition reshape and incremental advanced packaging/HBM demand relevant to TSMC and Korean memory makers.
Open source articleOriginal: 传高通洽谈为字节跳动提供定制芯片设计服务,目标年底量产VPU - 电子工程专辑
Chinese media reports Qualcomm is negotiating to provide custom chip design services to ByteDance for a video processing unit (VPU), with mass production targeted by year-end. The framing highlights how Chinese hyperscalers continue tapping US design houses for custom silicon despite export tensions, signaling sustained demand for advanced design IP and foundry capacity. Most affected in our universe: QCOM (new service revenue stream) and TSMC (likely foundry partner for the VPU).
Why it matters: Custom VPU deal between QCOM and ByteDance is a meaningful design-services win and likely TSMC foundry order, but it's a single-customer ASIC rather than a market-moving capacity or export-control shift.
Open source articleOriginal: 歐盟加入美國主導「矽盛世」倡議,降低依賴中國 AI 供應鏈
The EU, Netherlands, Germany and Greece signed onto Washington's 'Pax Silica' initiative, a non-binding pact to exclude China from AI-capable semiconductor supply chains spanning chips, critical minerals and energy. Argentina, Chile, Costa Rica, Kazakhstan and Panama are also joining this week, bringing membership to 24 countries — reinforcing the Western bloc's push to decouple AI hardware sourcing from China.
Why it matters: Geopolitical supply-chain framework with no named beneficiaries, capex figures, or binding commitments — sector-level signal rather than a stock-moving catalyst.
Open source articleOriginal: 9月华为一挑三!麒麟2026同场对决苹果A20、高通骁龙8 Gen 6和联发科天玑9600 - 驱动之家
Chinese media frames Huawei's September launch of the Kirin 2026 SoC as a one-vs-three showdown against Apple's A20, Qualcomm Snapdragon 8 Gen 6, and MediaTek Dimensity 9600, signaling SMIC-fabricated domestic silicon is now positioned as a peer to leading-edge foreign chips. The narrative reinforces China's self-sufficiency drive and implies share risk for Qualcomm and MediaTek in China smartphones, while a credible SMIC node advance pressures TSMC's China revenue mix.
Why it matters: Huawei Kirin re-entering flagship competition signals continued SMIC-led domestic substitution that pressures Qualcomm, MediaTek, and TSMC's China revenue, though impact is gradual rather than a single-event shock.
Open source articleOriginal: 엔비디아 시총 5조 달러 돌파 — 차세대 Vera Rubin 아키텍처 공개
Nvidia's market capitalization reached $5 trillion following the rollout of its next-generation Vera Rubin GPU architecture, succeeding the Blackwell platform. The milestone reinforces Nvidia's dominance in AI accelerators and signals continued strong demand for HBM, advanced packaging (CoWoS), and foundry capacity from TSMC, with downstream benefits for HBM suppliers SK Hynix and Samsung.
Why it matters: New flagship GPU architecture launch from Nvidia is a direct event driving demand visibility for HBM, advanced packaging, and TSMC foundry through 2026-2027.
Original: 日本擬砸 74 兆元投資 17 個領域,押注 AI 與半導體
PM Takaichi's government unveiled a roadmap to mobilize over ¥370 trillion (~NT$74T / ~$2.4T) of combined public-private investment through March 2041 across 17 strategic sectors, with semiconductors alone tagged at ¥68T, Physical AI at ¥10.5T and next-gen wireless at ¥20.5T. The framework will be embedded in the FY2027 budget under a new 'Strong and Prosperous Japan' investment vehicle, reinforcing Japan-led semi/AI capex demand that benefits upstream equipment, materials and foundry supply chains in Taiwan and Korea.
Why it matters: Long-dated national capex roadmap with no named beneficiary companies or contracts; directionally bullish for the AI/semi supply chain but not an immediate stock-moving catalyst.
Open source articleJul 14, 2026 close · day-over-day
MRVL
$218
-7.75%